Evaluating Defensive Performance Among The NHL's Defencemen

Steve Mitchell-US PRESSWIRE

A look at which defencemen are exceeding defensive expectations after a quarter of the 2013-14 season has been played with a focus on the Toronto Maple Leafs' defence

So over the off-season I (with the help of Frag) slowly morphed what had been SDI into a Multi-Variate regression that allowed us to compare Actual Corsi Performance for a player to Expected Corsi Performance now known as dCorsi. The expected results were the output from a regression that factors in a players' avg TOI, Zone Starts, and Weighted Average Team-mate and Opposition Corsi Results (both For and Against). Hopefully this stuff will be posted at ExtraSkater.com in the not too distant future. I've tabulated these results for Defenders AND Forwards at 5v5 Close, but I'm going to post the two groups separately. Today we'll start with the defenders.

I split up the expected Corsi For and Corsi Against results, as there is virtually no correlation between the two aspects of offensive and defensive play. I then explored the reliability of the statistic and came to the conclusion that 3 year data is pretty solid as a predictor going forward, and 5 year data is quite robust. The end result of all of this being that dCorsi does seem to paint a fairly accurate picture of how a player is performing offensively and defensively independent of the other guys on the ice. Unfortunately in small samples the results aren't guaranteed to last - sort of like how a goalie or shooter may get on a hot or cold streak and this isn't necessarily reflective of true underlying ability.

On the basis of the first 20-25 games it looks like P.K. Subban could well be on his way to another Norris type season.

Guys can have a number of solid games from a Corsi perspective, or even entire seasons that are good, but over the long term these things tend to smooth out. I just want people to bear this in mind as I post these results from the first 20 or so games of the season... this is an extremely small sample and it doesn't mean the players listed are future super-stars. BUT - that being said, it DOES reflect who is doing better (or worse) than expected based on their usage and line mates so far this year.

For the sake of overall comparisons, I'm going to post the listing of the Defenders by offensive and defensive results. dCF20 and dCA20. Note that the players with the lowest Expected CF20 have the tougher job on their hands to generate offense, and the players with the highest Expected CA20 have the toughest defensive assignments.

Defenders at 5v5 Close dCorsi For

Player Name Team 5v5 Close TOI TOI/GP CF20 ExpCF20 dCF20
P.K. SUBBAN Montreal 245.85 11.18 23.75 17.91 5.84
JARED SPURGEON Minnesota 228.40 11.42 21.19 16.68 4.51
MARK PYSYK Buffalo 206.92 9.00 16.05 12.08 3.97
THOMAS HICKEY NY Islanders 211.45 9.61 21.66 17.70 3.96
MATT NISKANEN Pittsburgh 211.87 9.63 21.71 18.03 3.68
CHRISTIAN EHRHOFF Buffalo 228.78 9.95 16.00 12.72 3.28
ANDREI MARKOV Montreal 229.08 10.41 22.44 19.17 3.27
TOM GILBERT Florida 220.37 10.49 18.97 16.12 2.85
JAMES WISNIEWSKI Columbus 231.52 10.52 16.76 14.17 2.59
DUSTIN BYFUGLIEN Winnipeg 282.93 12.30 21.70 19.27 2.43

You might need to get used to seeing P.K. Subban 's name a lot more often in Norris consideration. He's amazing. He was 2nd in the rankings amongst D overall over the prior 6 seasons to Zdeno Chara, but it looks like he's grabbed the mantle of top D man in the NHL fairly assertively... and it isn't even close.

Also - Mark Pysyk and Christian Ehrhoff are doing yeoman's work in Buffalo. Nobody else on the team seems to be doing anything positive, but those two are outperforming against massive odds. Those numbers look bad but they're being asked to do way too much in a horrible situation.

Defenders at 5v5 Close dCorsi Against

Player Name Team 5v5 Close TOI TOI/GP CA20 ExpCA20 dCA20
MARK PYSYK Buffalo 206.92 9.00 20.40 27.29 6.89
RON HAINSEY Carolina 236.18 11.25 18.80 25.15 6.35
CHRISTIAN EHRHOFF Buffalo 228.78 9.95 21.07 26.91 5.85
JAKE GARDINER Toronto 215.25 9.78 18.12 23.81 5.69
MARC STAAL NY Rangers 201.40 9.59 16.98 22.47 5.49
TJ BRODIE Calgary 244.07 11.09 18.85 24.11 5.27
MARC-EDOUARD VLASIC San Jose 222.58 10.12 15.72 20.94 5.22
RAPHAEL DIAZ Montreal 217.53 9.89 18.85 23.22 4.37
JOSH GORGES Montreal 235.02 10.68 18.55 22.72 4.17
JUSTIN FAULK Carolina 240.55 11.45 19.04 22.90 3.86

Remember how Ron Hainsey was available really cheaply as a UFA? Remember how Tom Gilbert was also? Maybe NHL GM's have no idea what they're doing? Just an idea - but when two of the guys that hit the open market and sat there for EONS end up in the top 10 rankings either offensively or defensively in this type of stat, someone somewhere isn't doing their job... actually a lot of someones.

Marc-Edouard Vlasic, Marc Staal, and Josh Gorges aren't really surprises on the defensive list - all 3 have played excellently, but Leafs fans will likely be shocked a tad by the fact that Jake Gardiner ranks 4th on this list. Again - look at what his Expected CA20 is - and then look at what his results are through 22 games. He's obviously not the problem on the blue line.

Also - Montreal pretty obviously has 2 offensive guys (Subban and Markov) and 2 defensive guys (Gorges and Diaz). This is news to absolutely nobody, but having the 4 of them rank in the top 10 in these two categories makes it look like good planning.

Top 10 Defenders in Overall dCorsi

Player Name Team CF20 ExpCF20 dCF20 CA20 ExpCA20 dCA20 dCorsi
MARK PYSYK Buffalo 16.05 12.08 3.97 20.40 27.29 6.89 10.86
CHRISTIAN EHRHOFF Buffalo 16.00 12.72 3.28 21.07 26.91 5.85 9.13
P.K. SUBBAN Montreal 23.75 17.91 5.84 19.61 21.39 1.78 7.62
MARC STAAL NY Rangers 22.15 20.91 1.23 16.98 22.47 5.49 6.72
RON HAINSEY Carolina 18.46 18.10 0.36 18.80 25.15 6.35 6.71
MARC-EDOUARD VLASIC San Jose 22.10 20.64 1.46 15.72 20.94 5.22 6.67
JAKE GARDINER Toronto 16.07 15.29 0.79 18.12 23.81 5.69 6.48
THOMAS HICKEY NY Islanders 21.66 17.70 3.96 17.03 19.01 1.99 5.95
ANDREI MARKOV Montreal 22.44 19.17 3.27 19.47 21.95 2.49 5.75
DUSTIN BYFUGLIEN Winnipeg 21.70 19.27 2.43 19.23 21.86 2.63 5.06

On the basis of the first 20-25 games it looks like Subban could well be on his way to another Norris type season. While his contribution assessed via dCorsi has him ranked 3rd consider the following:

1. his Norris win last year

2. His consistent results over his entire career so far (remember he ranked 2nd by this measure only to Zdeno Chara over the last 6 years)

3. The fact that Buffalo are the worst team in the NHL and there's no way in hell any journalist is going to name one of their D men as the best in the league.

I'm pretty sure if he can sustain his contributions he'll deserve any plaudits that come his way - oh and he should definitely be playing for Team Canada in Sochi at the Olympics this February.

Lastly I'd like to look at the Leafs blue line. Gardiner is obviously playing fairly well considering he's got the 7th best dCorsi in the NHL amongst D with 200+ 5v5 Close minutes so far. Let's see how the others rank:

Player Name 5v5 Close TOI TOI/GP CF20 ExpCF20 dCF20 CA20 ExpCA20 dCA20 dCorsi NHL Percentile
JAKE GARDINER 215.25 9.78 16.07 15.29 0.79 18.12 23.81 5.69 6.48 96%
PAUL RANGER 202.50 9.20 15.90 16.40 -0.50 19.85 22.68 2.83 2.34 68%
MORGAN RIELLY 152.08 8.95 18.15 14.92 3.23 23.28 21.90 -1.38 1.85 65%
CODY FRANSON 210.58 9.57 16.24 15.71 0.53 22.60 21.43 -1.18 -0.64 41%
CARL GUNNARSSON 202.85 9.22 14.79 18.02 -3.23 21.10 23.26 2.16 -1.07 36%
DION PHANEUF 214.00 9.73 16.54 15.19 1.35 23.18 18.81 -4.37 -3.02 23%

Well - this in a nut shell may explain much of the issues the Leafs are having overall. Cody Franson, Carl Gunnarsson and Dion Phaneuf aren't producing the required results to have the team win. Paul Ranger in particular is probably being bashed a tad more than he should be, as his results look fairly decent and place him in the top 3rd of NHL D men overall.

Most promisingly, Morgan Rielly's results make it fairly obvious he isn't out of place on an NHL blue line. Perhaps permanently playing him in the top 4, and reducing the load on Phaneuf, Gunnarsson and Franson by giving the other 3 more minutes would benefit the team as a whole.

This is also an excellent time to point out an interesting feature of this statistics. If you look closely at Gunnarsson and Phaneuf's Expected results in both CF and CA, you'll notice significant differences. Gunnarsson's expected CF is a lot higher than Phaneuf's because Gunnarsson plays with a good offensive player in Phaneuf while Phaneuf plays with a weaker offensive player in Gunnarsson. Thus Gunnarsson under-performs expectations offensively, while Phaneuf over-performs. This statistic lets you see who drives the offensive results in terms of shot attempt production.

When it comes to Expected Corsi Against, the situation is reversed. Gunnarsson's Expected CA is far higher than Phaneuf's because Gunnarsson is playing with the defensively weaker Phaneuf and Phaneuf is lined up alongside the reliable Gunnarsson. In the end, it appears Phaneuf isn't meeting expectations defensively, while Gunnarsson is holding his own. Again - it becomes clear which of the two is doing the tough part of limiting shot attempts against.

This should make intuitive sense and help explain the results you're observing.

The next posting in this series will provide the 1st quarter results for NHL Forwards at 5v5 Close, and hopefully lead to some interesting discussion.

Log In Sign Up

Log In Sign Up

Forgot password?

We'll email you a reset link.

If you signed up using a 3rd party account like Facebook or Twitter, please login with it instead.

Forgot password?

Try another email?

Almost done,

Join Pension Plan Puppets

You must be a member of Pension Plan Puppets to participate.

We have our own Community Guidelines at Pension Plan Puppets. You should read them.

Join Pension Plan Puppets

You must be a member of Pension Plan Puppets to participate.

We have our own Community Guidelines at Pension Plan Puppets. You should read them.




Choose an available username to complete sign up.

In order to provide our users with a better overall experience, we ask for more information from Facebook when using it to login so that we can learn more about our audience and provide you with the best possible experience. We do not store specific user data and the sharing of it is not required to login with Facebook.